> > The current proposal leaves most of the interesting use-cases on the table > rather than aiming to show that the abstraction actually meets the > requirements. >
I think quite the opposite. I think we will have a number of "interesting use-cases" implemented as ready-to-use timetables so in vast majority of cases you will have a simple, predefined "Timetable" implementation that will have at most some simple configuration, and it will be even better and simpler to use in "Scripting Ergonomics" and cover vast majority of cases. And being able to define your own aptly named schedules and use them in DAGs rather than copy & paste complex set of parameters to set in multiple Dags that want to reuse the same "kind" of schedule seems "ergonomically superior" by all means. J.
